clock menu more-arrow no yes mobile

Filed under:

Georgia Tech Football: WCU Advanced Stats Review - Defense

There were some numbers to love and some numbers to scare the GT faithful on Saturday night

COLLEGE FOOTBALL: SEP 10 Western Carolina at Georgia Tech Photo by Austin McAfee/Icon Sportswire via Getty Images

Final Score: Georgia Tech 35-17

Model Prediction: N/A

Projected EPA (Offense and Defense) Margin of Victory: GT by 22

GT Win Probability (Based on Success Rate, Yards Per Play, and EPA): 76%

Georgia Tech entered its game against Western Carolina with a win expectancy percentage somewhere in the lower 90s, based on Vegas expectations and models like SP+ that include FCS opponents. The actual play by results indicated something more like a 76% GT win expectancy, which echoes the frustration that many GT fans felt watching the game Saturday night. Against a team that lost its only FBS game in 2021 (to Oklahoma) 76-0, Georgia Tech could have and should have been more dominant. The height of the frustration came when the Catamounts scored easily on GT on not just their first possession but their second as well. Let’s dive into the numbers, with a focus on the defensive side of the ball for Tech.

Advanced Stats Comparison

When Western Carolina Had the Ball

Akshay covered the offense yesterday, and we’ll be focusing in on the GT defensive performance against Western Carolina (which will be our normal rhythm going forward this season). That means you can pay especial attention in the table above to the far right column, which displays the percentile performances for each metric for the WCU offense. It’s definitely not all bad news, but there’s enough above-average numbers for the Catamounts to raise some concern.

Rushing Defense

Against the run, Georgia Tech did a very good job of preventing any explosive plays but a not so good job of making WCU inefficient on the ground. On the one hand, the EPA/rush number is excellent, which is mostly driven by the lack of explosives and by GT’s above average run stuff rate. It was everything in the middle that was concerning. Two metrics to pay attention to here, Opportunity Rate (% of runs for four or more yards) and Average Line Yards (attempt to estimate the average yardage created by the OL on rushing plays), show that Western Carolina was able to succeed at above average rates on the ground.

The starting defensive line generally did their jobs on run plays, although the rush end position (mostly Collins and Kennard) had its struggles with gap integrity. Mostly though, players like Douse, Chimedza, White, and Bigger stood their ground, got penetration when needed, and set up the linebackers to clean things up. That’s where the intrigue lies.

Here’s the challenge facing the GT coaches: Trenilyas Tatum was GT’s highest graded defender (according to PFF) on run plays, while Ayinde Eley was its lowest. However, Tatum had one huge bust when he should have covered a running back on a third and ten play. Eley is a senior, a leader, a well-like player by all accounts. But he may not be the best option at Mike linebacker. He missed twenty percent of his tackle attempts on Saturday, and unfortunately that’s a trend, not an outlier. Better offenses will turn both kinds of these busts into much bigger plays than WCU was able to do; Zach Evans is heading to Atlanta on Saturday just waiting for those kinds of opportunities. Now that Charlie Thomas is back from his targeting suspension, the GT coaches have to figure out the right balance of snaps at the Mike spot. Let’s see what those snap counts look like this weekend.

Defensive Disruption

GT Disruption vs. WCU

Player Defending Havoc Plays # of Pressures # of Run Stuffs PFF Grade
Player Defending Havoc Plays # of Pressures # of Run Stuffs PFF Grade
Sims 1 0 0 72.8
Thomas 3 2 1 69.7
Eley 2 6 1 59.6
White 1 3 0 61
Scott 1 3 1 53.1
Kennard 1 3 2 50.7
Tatum 0 1 0 73.7
Collins 0 0 1 55.2
Douse 0 1 0 73.7
Stone 0 2 0 72.9

Georgia Tech got off to a slow start disruption wise but ended up posting pretty good numbers in the pressure department. Carlos Davis, WCU’s starter at quarterback who was injured and left the game in the third quarter, posted a -0.50 EPA/drop back when facing pressure versus -0.10 EPA/drop back with no pressure. As the Georgia Tech defense began to find more disruption, the effectiveness of the WCU offense dropped accordingly.

Looking at the individual player numbers, the Ayinde Eley conundrum stands in stark relief here; the 6 pressures he generated is a fantastic number, but his overall numbers don’t look great because of his struggles finishing plays and maintaining proper leverage. He wasn’t the only one; Scott and Kennard were responsible for 3 pressures each but had below-average grades because of the opportunities they gave to WCU quarterbacks on those pressure plays. Keion White was far and away the best defensive lineman most of the time, but his multiple offside penalties hurt his overall metrics. That was a bizarre occurrence, but White’s 7 pressures and 4 sacks through two games are a huge encouragement for GT’s ability to continue finding some level of disruption against opposing passing games. Unfortunately, consistency and awareness from the rush end spot are still lacking, and there is a clear drop-off from former Jacket and current Razorback Jordan Domineck to the two guys playing most of the snaps so far.

Pass Coverage

GT Coverage vs. WCU

Player in Primary Coverage Targets CPOE Allowed YAC Allowed PFF Grade
Player in Primary Coverage Targets CPOE Allowed YAC Allowed PFF Grade
Sims 1 -53.00% 0 72.8
Thomas 1 -69.67% 0 69.7
Eley 1 22.50% 3 59.6
Allen 1 -37.00% 0 54.8
King 2 46.90% 6 67.3
Brooks 1 25.00% 4 64.2
Wallace 3 12.90% -1 47.2
Kennard 1 27.67% 4 50.7
*Target numbers here are incomplete, as I was not able to verify the individual target charting on many plays because I didn’t have rewind capabilities while watching. Blame ESPN+.

The coverage busts on the first two drives for WCU were eerily reminiscent to last year, but thankfully, those issues improved as the game wore on. The cornerback play has been much more consistently reliable in the first two games. Of course, there are still a few issues of concern here.

There was a schematic decision to regularly deploy Kyle Kennard in coverage (PFF has him targeted 5 times), and that didn’t go well, as four of those were completed and two included missed tackles from Kennard. Eley was also quite ineffective when asked to cover from his linebacker spot. Lane Kiffin is likely to try and take advantage of match ups like that with his excellent offensive weapons.

Like last year, there are personnel questions here that are confusing me. Sims, Walton, and King all graded quite well and avoided any obvious busts. They clearly deserve the bulk of the snaps at their respective positions. For the second straight week, LaMiles Brooks seemed to outplay Derrik Allen, and it’s a complete mystery to the writer why KJ Wallace played every defensive snap on Saturday while Kaleb Edwards played none. I would love an answer, which I probably won’t get, so I will settle for a change in that deployment pattern.

EPA Highlights

EPA calculates the expected number of points added (or lost in the case of a negative number) on a particular play based on the down and the location on they field.

As always, we’ll take a look at the most helpful and hurtful plays for GT.

Most Helpful Plays

  1. 5.68 EPA: Charlie Thomas’s interception on 2nd and 11 in the fourth quarter, returned to the WCU 12.
  2. 5.30 EPA: Georgia Tech’s recovery of a fumbled WCU snap on 3rd and 1 from the GT 36, recovered at the WCU 46.
  3. 4.53 EPA: Dontae Smith’s 51 yard touchdown run on 2nd and 1.
  4. 4.08 EPA: Myles Sims’s interception in the second quarter, returned to the WCU 24.
  5. 3.85 EPA: Nate McCollum’s 40 yard touchdown run on 2nd and 9.

This pattern looks quite different from the Clemson game. Instead of Jeff Sims accounting for three of the four most impactful plays as he did last week, the WCU highlights include three turnovers and two runs. That gives cause for concern.

First, the 5.30 EPA that GT gained because of a botched Western Carolina snap was 100% luck and not in any way predictive of future outcomes. That one play constituted about 25% of GT’s EPA margin in this game. Overall, per Game on Paper, GT experienced almost 22 points of turnover luck in this game! Sure, those turnovers were important within the game itself, but what turnover luck is trying to tell us is that those plays are not predictive of future performance. Balls bounce differently, and this could have been an even tighter game for GT. The absence of any passing highlights is quite concerning, as Akshay covered yesterday.

Most Hurtful Plays

  1. -3.95 EPA: WCU’s 49 yard touchdown pass on its opening drive.
  2. -3.49 EPA: Jude Kelley’s missed 24 yard field goal in the fourth quarter.
  3. -3.17 EPA: Jeff Sims’s interception in the third quarter, returned to the WCU 16.

On the one hand, Georgia Tech did a very good job limiting WCU explosives after the first ten minutes of the game. That’s born out by the lack of big EPA plays for the Catamount offense as the game worn on. On the other hand, the missed field goal was deflating, given the years-long struggle for GT in this area, and the Jeff Sims interception was mind-blowing. Not facing any pressure, he threw a deep ball in a place where it seemed the only person who could catch it was the defender instead of the other way around. The receiver situation isn’t helping the lack of passing explosives, but this is a game where Sims should have been able to hit explosives and play turnover-free, given the lack of talent on the opposing defense. Instead, the concerns from the past two years continue.

Tracking Season Goals

*I set these goals for the 2022 season in some of my offseason preview work. We will be tracking them as we go this year.

GT Season Goals vs. WCU

Metric Season Goal This Week Season Long
Metric Season Goal This Week Season Long
GT CPOE >= 2% -11% -4%
Pressure Rate Allowed <=26% 6% 20%
Run Rate on 2nd Down and Long <=40% 78% 64%
Average Depth of Target >=9 14 10
Defensive Passing EPA/play <= 0.08 -0.04 0.04
Defensive Havoc Rate >=18% 16% 16%
Defensive Pressure Rate >= 27% 24% 22%

The passing ineffectiveness is my biggest concern here. I wish the second down and long run rate was much lower, but those decisions seemed directly related to the struggles Sims showed in the first half when he got more chances to throw. We got the deeper average target depth we wanted to see, but it wasn’t effective at all. Sims is yet to complete a pass of 20 or more yards down the field in two games.

Defensive havoc and pressure rates aren’t quite where we want them, but there seems to be clear improvement from last year in those areas of disruption. As many have observed, this stretch of three games against Ole Miss, UCF, and Pitt is going to tell us plenty.


  1. The defense looks more disruptive, but we need a bigger sample size to know for sure. Keion White and Charlie Thomas are players, but it’s not clear who else can be consistent to go with them.
  2. The defense does not reliably finish plays. Missed tackles, poor rush lanes, and failures to set the edge against the rush will likely lead to many more explosives in the coming weeks if not cleaned up.
  3. There are a few key personnel decisions that need to be made. At one linebacker spot, one safety spot, and the nickel back spot, the results so far call into question the snap distributions.
  4. The degree of difficulty for the defense is about to ramp up. The next three offenses GT faces are 11th, 39th, and 10th according to SP+. What does the defense look like after this crucial stretch? We’re about to learn a lot.